Building Better Models

with JMP Pro

Companion Site

This is the textbook support page for Building Better Models with JMP Pro, by Grayson, Gardner and Stephens, SAS Press, 2015.

The data sets for the case studies used in the book, along with some scripts used for illustration, can be downloaded from the SAS Press download page.

Note that some data sets are in the JMP Sample Data Library. Links to course materials developed by two professors are provided below. Errata, which will be addressed in the next edition, are listed here as well.

Links to Useful Resources:

Live and recorded webinars for getting started with JMP, data analysis, graphics, data preparation, and modeling.
JMP Case Study Library: Business-oriented and analytics case studies, from basic graphics to multiple linear and logistic regression, classification trees, neural nets, and model validation and selection.
JMP Learning Library: One page guides and short videos on a number of topics.

Course Materials:

Jim Grayson teaches a graduate level "Business Analytics for Managers" course at Augusta University. His course materials, including his schedule and syllabus for a previous course, are available for download here.

Download

Bob Nydick teaches a graduate level "Analytical Methods for Data Mining" online course at Villanova University using this book. Materials he has developed for the course (not including Multiple Regression and Logistic Regression), are available for download here.

Download

Errata and Revisions:

Page 48 - Step 3 should read “Click the carat (^) insert symbol in the formula editor….“
Figure 4.24 - needs to be retaken with Log(Bal_Total) instead of Bal_Total.
Figures 4.26 and 4.27 need to be retaken with AccountAge.
Page 88 - change "11 remaining" to "12 remaining."
Page 117 - The description of the markers is wrong. "Yes" is marked with an "o" symbol and "No" is marked with a "+" symbol.
Chapter 5, remove Exercise 5.7.
Figure 6.10 is incorrect.
Figure 6.12 is incorrect (the one shown is after 6 splits).
Bottom of page 158 - last sentence, missing parenthesis. It should read, "... has 15 splits (see Figure 6.26). Note..."
Figure 7.22: Add a note on page 202, results will be slightly different on a Mac (using the same random seed).
Figure 7.33 and 7.34: The output will be slightly different on a Mac (using the random seed of 1000).
Page 218, Exercise 7.1. Should be changed to: In the example, we fit a neural network with one layer and three notes, all using the TanH activation function.
a. Fit the model described in Example 1, and also fit a model with one layer with three nodes, using TanH, linear, and Gaussian activation functions.
Page 232, first sentence in the example should read:
"...to illustrate how cross-validation is used in building predictive models."
Page 234: Add the following sentence, "Note: Rows are randomly assigned to the training, validation and test sets. So, your validation column will likely be different than what is shown here. Use Boston Housing BBM Ch8 with Validation.jmp to produce the results shown in the remainder of this chapter."
Page 251 and 252, exercise 8.1 and 8.2: the data set is Boston Housing BBM Ch8.jmp.
Page 253, exercise 8.5: The Credit Card Marketing BBM.jmp data is described in Example 1 in Chapter 6.
Figure 9.6 - the caption should read "Boosted Tree Prediction formula for Prob[Survived=Yes]."
A professor pointed out that in the dataset Churn.jmp there are spaces in front of the values in all of the Nominal columns. For example, for the variables IntlPlan and VMPlan the values are " yes" and " no", rather than "yes" and "no." This isn't a problem for modeling, but if you save the prediction formula to the data table and then enter a new row, you need to enter the values with the space in order for JMP to make a prediction. An alternative is to first recode the columns to remove the space. This is easily done by selecting the columns in the data table, and using Cols > Utilities > Recode - click on the top red triangle in Recode and select Trim Whitespace.
A professor requests including a discussion of the FDR (False Discovery Rate) checkbox and what it means in the next edition of Building Better Models with JMP Pro. Noted!

Note: On 12/7/2015 an updated version of the data sets was uploaded to the SAS textbook page. The following changes were made:

The missing value column property was removed from the CreditRiskModeling BBM.jmp data table (used in Chapter 7).
The CreditRiskModeling BBM Clean.jmp was added. This file has been prepared for modeling (see page 211).